The authors present suggestions for selecting psychoanalytic candidates, derived from a study of psychiatric residents at the Menninger School of Psychiatry. Predictions from interviews alone proved to be slightly better than chance predictions. If the interviews are supplemented by a battery of tests, their predictive value is greater. One error in interviews and tests alike is to overemphasize the pathology and pay too little attention to the strengths and the history of previous functioning. Results of stress interviews may be misleading, for a person's functioning in a passive, reactive relation to a threatening authority may be very different from his reaction to being in control as is usual in his work. A battery of tests should be used, and administered by people who, by experience, have proved to be good interpreters of function in psychiatrists. Holt and Luborsky offer suggestions for keeping simple records of interview impressions during the candidate's training to check the validity of selection techniques.